
************************************************************************************************
Overview
************************************************************************************************

This replication package contains the data and the code necessary to reproduce the results in the paper “The Selective Disclosure of Evidence: An Experiment” (A. Farina, G. Fréchette, A. Ispano, A. Lizzeri, J. Perego).


************************************************************************************************
Data Availability Statement
************************************************************************************************

The data used in this paper were collected by the authors via a laboratory experiment conducted at New York University and Columbia University in 2023. 

The dataset is included in the replication archive in the folder replication_main/data. 

Documentation of the experimental and sampling procedure is provided in the file replication_main/experiment_documentation/additional_documentation_experiment.pdf. This file contains a description of the experimental interface and experimental instructions, it summarizes the subjects' recruitment procedure and the data exclusion criteria, and it provides a description of the variables in the raw data and of the variables after the data cleaning. 

The data are made available under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.




************************************************************************************************
Computational Requirements
************************************************************************************************

******************************
Software Requirements
******************************

This replication package uses three software environments:

	1	Python / Jupyter:
		- Anaconda - Navigator (Jupyter Notebook)
		- Python 3.10
		- The full Python environment is specified in environment_restud.yml, with the version of the packages last used to run the code
	
	2	Stata:
		- Stata/SE 17.0
	
	3	Matlab:
		- Matlab_R2024_a


****************************************************
Instructions to Create the Python Environment
****************************************************

Open the Terminal and set the correct path (depending on where the folder is unzipped): cd "..."

Then run:

conda env create -f environment_restud.yml


The Python code was tested using this environment.



**********************************************************************
Folder Structure with explanation (in alphabetical order)
**********************************************************************

replication_main/

	analysis_main.do:
		file containing all the analyses performed on Stata. In the file, it is explicitly stated which table and figure each section produces

	data:
		folder containing raw data (in csv format) and intermediate files

	dat_fflp_1_compact_complete.dta:
		final dataset used for the analyses

	dat_fflp_1_compact.dta:
		intermediate data file, only used to generate dat_fflp_1_compact_complete.dta

	data_fflp_1.do:
		do file generating dat_fflp_1_compact.dta from the raw data

	data_frame.ipynb:
		Python notebook generating dat_fflp_1_compact_complete.dta from dat_fflp_1_compact.dta

	environment_restud.yml:
		Conda environment specification for the Python replication environment

	experiment_documentation:
		folder containing the additional experimental documentation mentioned in the Data Availability Statement section

	Figure2(a)-2(b)-2(c)-2(d)_SenderCDF.ipynb:
		Python notebook generating Figures 2(a)-2(d). The second part of the notebook computes values that are then imported into policy_simul.m to perform the simulations reported in Section 5.1

	Figure3(a)-3(b)-C6-C7-C8-C9_SenderCluster.ipynb:
		Python notebook generating sender clustering and Figures 3(a)-3(b)-C6-C7-C8-C9

	FigureC5(a)-C5(b)_SignalDistribution.ipynb:
		Python notebook generating FigureC5(a)-C5(b)

	FigureC10-C11_ReceiversGuesses.ipynb:
		Python notebook generating Figures C10-C11

	FigureC12(a)-C12(b)_ReceiverGapCDF.ipynb:
		Python notebook generating Figures C12(a)-C12(b)

	figures:
		folder where all generated figures are saved when the code is run. This folder also contains the subfolder do where Stata graphs used in the final figures are saved

	policy_simul.m:
		Matlab code generating the policy simulations reported in Section 5.1

	README.txt:
		this file

	Refinement_Assumption_SessionLevel.ipynb:
		notebook checking refinement assumptions at the session level

	Refinement_Assumption_TreatmentLevel.ipynb:
		notebook checking refinement assumptions at the treatment level

	Robustness_Correlation_Analysis_Receiver.ipynb:
		Python notebook performing robustness analysis of receiver informativeness

	Robustness_Correlation_Analysis_Sender.ipynb:
		Python notebook performing robustness analysis of sender informativeness



*************************************
Instructions to Run the Code
*************************************

Order of files:

In order to perform the analysis, the first file to run is data_fflp_1.do, which creates the intermediate dataset dat_fflp_1_compact.dta. After this file, the second one that needs to be run is data_frame.ipynb, which creates the final dataset dat_fflp_1_compact_complete.dta. Once this dataset is generated, all the other analyses can be performed in any order.

comnd
Python / Jupyter:

In order to run the .ipynb files, after creating the Conda environment as indicated above, the Jupyter notebook needs to be started in the following way:

	•	Open the Terminal and set the correct path (depending on where the folder is unzipped): cd "..."

	•	In the Terminal: activate the environment: conda activate restud_replication

	•	In the Terminal: python -m notebook (the jupyter interface should open automatically, but if not, you should see a URL that you can copy and paste in the browser)

	•	Open the notebooks in the replication_main folder, if needed set the kernel's working directory to the "replication_main" folder, and run all cells sequentially



Stata

In order to run the .do files:

	•	Open Stata/SE 17.0

	•	Set the Stata working directory to the "replication_main" folder/ execute the do files from the "replication_main" folder

	•	Run the .do file section by section (beginning and end indicated in the code for each section)


Matlab

In order to run the .m file:

	•	Open Matlab_R2024_a

	•	Set the Matlab's current folder to the "replication_main" folder

	•	Run the whole file: the informativeness results will be displayed in the Command Window with clear labels






************************************************************************************************
Notes
************************************************************************************************

	•	The replication package is fully self-contained

	•	All data required to reproduce the results are included in the data folder

	•	Manual path adjustments are necessary and clearly indicated. The beginning of each file specifies where the path adjustment needs to be made

	•	The Python notebooks were tested using the Conda environment specified in environment_restud.yml

	•	The code was tested on macOS

	•	The font of the figures may differ from the font in the paper, to avoid errors due to the possible unavailability of the Times New Roman font

	





************************************************************************************************
Contact Information
************************************************************************************************

For any issue or question about this replication package, please contact:

Agata Farina
agata.farina@chicagobooth.edu



